Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.02178 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912870589005824 |
|---|---|
| author | Lin, Liang Xiong, Feng Wang, Zengbin Wang, Kun Dong, Junhao Hu, Xuecai Wang, Yong Chu, Xiangxiang |
| author_facet | Lin, Liang Xiong, Feng Wang, Zengbin Wang, Kun Dong, Junhao Hu, Xuecai Wang, Yong Chu, Xiangxiang |
| contents | Diffusion Large Language Models (DLLMs) have emerged as a powerful alternative to autoregressive models, enabling parallel token generation across multiple positions. However, preference alignment of DLLMs remains challenging due to high variance introduced by Evidence Lower Bound (ELBO)-based likelihood estimation. In this work, we propose AR-MAP, a novel transfer learning framework that leverages preference-aligned autoregressive LLMs (AR-LLMs) as implicit teachers for DLLM alignment. We reveal that DLLMs can effectively absorb alignment knowledge from AR-LLMs through simple weight scaling, exploiting the shared architectural structure between these divergent generation paradigms. Crucially, our approach circumvents the high variance and computational overhead of direct DLLM alignment and comprehensive experiments across diverse preference alignment tasks demonstrate that AR-MAP achieves competitive or superior performance compared to existing DLLM-specific alignment methods, achieving 69.08\% average score across all tasks and models. Our Code is available at https://github.com/AMAP-ML/AR-MAP. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_02178 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | AR-MAP: Are Autoregressive Large Language Models Implicit Teachers for Diffusion Large Language Models? Lin, Liang Xiong, Feng Wang, Zengbin Wang, Kun Dong, Junhao Hu, Xuecai Wang, Yong Chu, Xiangxiang Computation and Language Diffusion Large Language Models (DLLMs) have emerged as a powerful alternative to autoregressive models, enabling parallel token generation across multiple positions. However, preference alignment of DLLMs remains challenging due to high variance introduced by Evidence Lower Bound (ELBO)-based likelihood estimation. In this work, we propose AR-MAP, a novel transfer learning framework that leverages preference-aligned autoregressive LLMs (AR-LLMs) as implicit teachers for DLLM alignment. We reveal that DLLMs can effectively absorb alignment knowledge from AR-LLMs through simple weight scaling, exploiting the shared architectural structure between these divergent generation paradigms. Crucially, our approach circumvents the high variance and computational overhead of direct DLLM alignment and comprehensive experiments across diverse preference alignment tasks demonstrate that AR-MAP achieves competitive or superior performance compared to existing DLLM-specific alignment methods, achieving 69.08\% average score across all tasks and models. Our Code is available at https://github.com/AMAP-ML/AR-MAP. |
| title | AR-MAP: Are Autoregressive Large Language Models Implicit Teachers for Diffusion Large Language Models? |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.02178 |